Angelo Theodorou

As a freelance rigger, I need to make sure my builds have a lot of flexibility. So I've recently reworked my lid modules to add variable joint counts. By utilizing a curve-based output I can changed the amount of joints used based on the needs of the project. Higher joint counts for fidelity and control, lower counts for simplicity and optimization. Or I can test and iterate to find a happy middle ground. So animators using my rigs on film or game projects can enjoy the same features, such as: - Automated lid follow with eye movement - Main controls to position the lids through translation and sculpt the curvature with rotation/scale  - Secondary controls for finer detail (If there are enough joints to support it) - Manual or automated blinking. Animate controls, or activate the automated meet point between your posed lids for a perfect seal I've been using iterations of this system over my last few projects, both commercials and games. The core elements of this system have been floating around in my head since 2018, so it's been great to finally put it to the test. 

1,528

27 Comments

Just found this undocumented feature added in UE5.5+ which allows standard Editor Utility Widgets to be dockable as a main tab instead of requiring it to be nested. This is especially helpful for those making widgets that require larger real estate or embedded viewports. I was digging in C++ source the last little while to see how to achieve this but looks like it was added in a different, and more accessible form, for newer versions of the engine already.

20

My Python Script Editor plugin for Unreal has been updated to v2.1.0 This adds support for UE-5.6 and now supports unlimited scrolling tabs so you can bring all your bad habits over from Maya. Also thanks to Mateusz it now compiles with -StrictIncludes enabled. You can get it here on Fab: https://lnkd.in/ghSGgxga Also to answer some questions I've had: • Yes this uses a custom slate tab widget, not the built in one. • Yes you can use the tab widget components in your own projects and products, I'm not that attached to it, I know the built in FTabManager is a pain to work with. • Yes the Editor is designed to work across projects/users, just set the preferences to store tabs in your documents folder. • No I'm not planning to add AI autocomplete. • Yes I agree Unreal should come with a script editor, it doesn't which is why I wrote this. I'd be happy to chat if Epic want to take it over. #unrealengine #ue5 #slate #python

114

4 Comments

Following up on my previous post about using Modular Rig for simple environment animation, but this time focusing on a mechanical prop. This prop uses slightly more complex module than the previous one, and the results were honestly a bit surprising (in a good way). ➤ What worked well ✅ Moving beyond single-joint motion is doable, as long as the prop parts are spitted with animation in mind. ✅ Building a skeleton and painting skin weights directly in Unreal works well for this prop, which has a moderate complexity. ✅ Mechanical modules like Hinge and Piston are already well supported out of the box and easy to reason about. ➤ Where friction shows up ⚠️ This still isn’t a full DCC replacement. The quality of the result depends heavily on how well the asset is designed and segmented upfront. ➤ Bigger takeaway For me, this reinforced that Modular Rig lowers the barrier to keeping work entirely in-engine and expands what environment art and level design can include, especially for smaller projects where DCC round-trips aren’t available or are too costly. On top of that, it could potentially enable iteration to happen directly in the level. Still exploring, but I’m curious how others are thinking about where this kind of workflow fits into their pipelines. 🤔 #UnrealEngine #UE5 #TechnicalAnimation #GameDev #TechnicalArt #LevelDesign #ControlRig #AnimationPipeline

352

13 Comments

One of the main reasons many people hesitate to use Resolve in a semi-professional restoration context is the lack of control over its automatic tools. To address this, I introduced an enhanced version of the automatic dirt removal tool at the IASA/SEAPAVAA Conference in Hawai‘i, now called Advanced Dirt Removal. This version introduces three key improvements: Recovery Brush : By combining the tool with a Mask Paint node, you can restore any artifacts introduced by the automatic process. Motion Mask Protection : Using motion vectors, the motion mask safeguards high-motion areas from being over-processed, preserving detail where it matters most. Transparent Difference Mode : The image can be displayed in a semi-transparent difference view, clearly showing what has been removed and allowing for quick and precise recovery when needed. Additionally, with a simple toggle you can switch Advanced Dirt Removal into “Spanish Mode.” In this mode, the effect applies only to your brush strokes, ensuring maximum quality with no wasted time hunting for the correct frame to paint, and minimizing artifacts by giving you complete control over the process. I will be presenting this tool again at the AMIA Conference in Baltimore as part of the updated version of my workshop. I would love to hear your suggestions or feedback, as with my other projects I am always looking for ways to refine and improve functionality. #FilmRestoration #ArchivalHeritage #MediaPreservation #PostProductionTools #AMIA2025 https://lnkd.in/e83WmiNm

37

3 Comments

 I gave Claude Code the ability to debug GPUs.  I built a skill that connects Claude Code to RenderDoc — the industry-standard GPU debugger. Now when I describe a rendering bug, Claude can:  - Launch my app and capture a GPU frame  - Inspect pipeline state at any draw call  - Export render targets and actually see what the GPU rendered (Claude is multimodal)  - Trace shader execution line-by-line  - Debug individual pixels to find out exactly why they look wrong  In the demo, I ask Claude to capture a frame from a Vulkan renderer and figure out why the sphere is black. It captures the frame, exports the render target, inspects the vertex shader,  and discovers the root cause: the MVP matrices are all zeros because the frame was captured before uniform buffers were initialized. Full diagnosis in under a minute.  This works through rdc-cli by Jim (BANANASJIM) — a brilliant 66-command CLI that wraps RenderDoc's Python API into shell commands. I wrote the Claude Code skill layer on top: the recipes,  the visual inspection patterns, the debugging workflows. Claude learns how to use all 66 commands and when to apply each debugging recipe.  Works with Vulkan, D3D11, D3D12, and OpenGL. Open source (MIT).  GitHub: https://lnkd.in/gav5yXNr  Video: https://lnkd.in/gat9QCaF  #ClaudeCode #GPU #RenderDoc #Vulkan #GraphicsProgramming #AI #GameDev #OpenSource

440

23 Comments

Hey, if it ain't broken... Clair Obscur: Expedition 33 studio doesn’t plan on expanding its scale of development. “We’re just the right size for a full-price turn-based RPG” Expand as needed. Don't expand for expansion sake Clair Obscur: Expedition 33 became one of 2025’s most acclaimed games, with many praising its development approach. Despite its success, Sandfall Interactive isn’t planning to scale up. “For now, I’d prefer working as a small team… when it comes to making a full-priced turn-based RPG, I believe that the team we have now is just the right size,” - Tom Guillermin. Creative director Guillaume Broche emphasized that although QA, localization, and marketing were outsourced, the core team was small: “We had five people working on environments, two on the story… music was done by four people.” Guillermin noted that some members were fresh graduates, which allowed them to adapt easily to the studio’s unique workflow. I’ve been in gaming long enough to watch studios grow chasing the next spectacle, only to collapse under their own weight. There’s something powerful about a studio that knows its limits and embraces them. Instead of scaling up just because they found success, Sandfall is doubling down on what worked: a tight, artistically driven team making exactly the kind of game they believe in. In E33, they were able to focus on what matters most and do a small amount of things really well rather than do a ton of things that are just alright. We’ve been conditioned to equate size with value. But Expedition 33 proves that restraint can actually enhance quality. While I would have welcomed a longer runtime, I was grateful to play something polished, beautiful, and complete. Games like this remind me that focused titles aren’t just viable, they might be what the industry needs most.

136

10 Comments

I want to demystify how ASGC tracks games industry job cuts so you understand what goes into the numbers The way we track industry reductions is far more rigorous than most people realize. I occasionally see comments wondering if we only use public announcements, WARN notices, or that we miss contractors, co dev teams, and non traditional roles. I understand why people might assume that, but that is not how our system works. Collecting jobs data is the most resource intensive process in our community because it shapes our support too. Step one is public information. We monitor news reports, company statements, WARN filings, and government notices. That is just the starting point. Step two is direct community reporting at scale. Every year I receive 5-10K+ messages, just related to job cuts, across LinkedIn, Discord, and email from people sharing what happened to them, their teammates, or their organizations. Much of this never appears in the press. It includes contractors, co dev partners, support studios, and indirect roles, not just FTEs at major publishers. Step three is individual signal tracking. I regularly review posts from professionals who are suddenly open to work or referencing team changes. These signals confirm patterns or reveal cuts that were not announced. All of this flows into a large internal tracking system that helps me understand not just how many roles were affected, but who was impacted, where, and when. That context allows our community to reach out and design support that matches reality rather than headlines. Is it perfect? No. I am still working to improve visibility in regions where transparency is lower. But I can say sincerely this process goes far beyond public records and is more comprehensive than any single external source. You may notice that, for a few years, I no longer publicly name organizations when cuts happen unless they request a Games Org support post. That is intentional. My goal is to avoid errors and avoid shaming, because I recognize that not every reduction comes from bad intent. Sometimes funding ends, contracts fall through, or projects conclude despite leaders trying to do right by their teams. Also, my approach is to build bridges and encourage quiet, meaningful accountability. I regularly hear that organizations know ASGC tracks layoff decisions and understand their actions will reach me, even with smaller or less visible cuts. That awareness, even without public callouts, can influence how situations are handled. Most importantly, workers know their experience is not invisible. Individual data is never shared publicly. People trust me with personal information, and I take that seriously. What I share are patterns, totals, and insights that help us mobilize support. So when we talk about industry cuts, please know this is not casual scorekeeping. It is the result of thousands of conversations and careful tracking to make sure every affected games person counts.

29

3 Comments

Quick trick to cut the bandwidth usage of your online game by 87% when sending status effects ("Poisoned, stunned, etc.."): Store all status effects in a bitmask instead of using a boolean for each of them. "bool stunned", "bool slowed", "bool poisoned" -> take 1 byte each. "byte statusEffects" -> takes 1 byte in total [0000 0000], each bit represents a specific status effect (up to 8) If the value of the bit is "1", you are affected by that state. For example: "Poisoned" is the right-most bit -> its integer value is "1" "Slowed" is the bit before it-> its integer value is "2" You're Poisoned and Slowed? then your statusEffects' value is 3 => [0000 0011] in binary Want to know if you're Poisoned? Just use a bitwise AND with the value of the status: If (statusEffects & 1) == 1 --> you're slowed If (statusEffects & 2) == 1 --> you're poisoned If your game has up to 8 status effects, this trick will save you up to 7 bytes on each character/npc instance, which makes a lot of difference at scale (you pay 87% less for bandwidth, and encoding/decoding packages requires less CPU!) #JustDevIt  #gamedevelopment #multiplayer #netcode #optimization

98

36 Comments

The Production version of my physics simulation tool is now on FAB! Will submit the base version later today. Update includes + Re Instance, using stored transforms from other sims + Distance to Center + Merge runtime actors + Convert to static Mesh https://lnkd.in/dGmywPkP Will be updating Documentation asap. There is a lot to add with more in depth and detailed examples. Erase function is temp disabled while fixing slate keys will sort this in the next few days.

31

2 Comments

👨‍💻 Ever wondered how `dict` or `HashMap` works under the hood? This week, I took a step back from high-level abstractions and built a hash map from scratch — just buckets, hashing, and a little resizing magic. Most languages give us a hash map for free: Python: `dict` JavaScript​: `Object` / `Map` Rust: `HashMap` 🦀 Go: `map` But that performance we take for granted relies on a careful balance of: ✅ Good hashing ✅ Collision handling (I used separate chaining) ✅ Resizing based on load factor ✅ Rehashing when capacity changes It's easy to forget just how much is going on inside that simple `map[key]`. ✍️ I wrote a blog post about it — complete with diagrams and a Python implementation: 🔗 https://lnkd.in/gz3UehiG Whether you're new to systems thinking or just love peeling back the layers of abstraction, I hope you find it helpful! #programming #datastructures #python #rustlang #buildslowthings #career #learning

22

3 Comments